Abstract
The article illustrates some of the strategies we are developing in the secondary analysis of Timescapes data and seeks to draw some general lessons for qualitative data analysts. We focus on three different areas of work. Across all of these we examine the potential explanatory value of working with data in a comparative way, and engage with some challenges presented by contextual specificity in the way qualitative data are generated. In the first area we consider the issue of how we situate qualitative data with reference to diversity across the population, and use an example of working between a single qualitative Timescapes data set and survey data. Understanding how qualitative data are situated offers a framework for internal comparison which maps onto wider diversity. In the second area we consider the outcome of bringing together primary researchers whose comparison of project data, as secondary analysts, allow them to ‘hear silences’ and, therefore, re-interrogate their own data within a revised conceptual framework. In the third area we describe how, as secondary analysts, we have worked across Timescapes data sets. Here we consider the challenges of undertaking secondary analysis across diverse, project specific, research contexts, and the potential of comparative working across data sets for enhancing understanding.
Keywords
Introduction
Secondary analysis of qualitative data is an area with significant potential for growth. In social science research the process is already well recognized, with growing numbers of publications and presentations on secondary analysis. Frequently, such work reports on the re-analysis of data held within a single data set (e.g. Fielding and Fielding, 2000; Savage, 2005), although secondary analysts also work across data sets (e.g. Bishop, 2007; Savage, 2010). As a practice it has been accompanied by debate about its conduct and effectiveness, and the challenges it presents (e.g. Gillies and Edwards, 2005; Hammersley, 1997, 2010; Moore, 2007). Qualitative data provide challenges very different to those inherent in quantitative data. Quantitative research is predicated on standardization whereas qualitative research is centred on specificity, context and the embedded nature of meaning. By its nature, then, it engenders particular difficulties for secondary users. However, the secondary analysis of qualitative data also holds a significant potential in enhancing the capacity of such evidence to contribute to building knowledge. In longitudinal research the challenges, but also potential, are magnified.
Within Timescapes the creation of an archive of digitized data sources will enable researchers to explore and analyse data which others have created. The Timescapes secondary analysis programme is centred on developing methodological, as well as substantive, insights. We discuss our philosophy and develop an argument for a critical secondary analysis elsewhere (Irwin and Winterton, 2011a, 2011b). We are undertaking a range of different activities as part of the Secondary Analysis strand of Timescapes. 1 These include developing strategies for working with primary data sets generated by the different project teams, and working across diverse data sets both within Timescapes, and between Timescapes and external data sets.
In this article we reflect on the potential explanatory value of working with data in a comparative way, and engage with the challenges presented by contextual specificity in the way data is generated. Issues of context loom large in discussions of the efficacy of secondary analysis (e.g. Hammersley, 1997, 2010; Mauthner et al., 1998; Moore, 2006). These tend to relate to presence at the point of data generation and linked knowledge of the proximate setting and of participants. With others (e.g. Bishop, 2009; Hammersley, 2010) we argue that while primary analysts have a privileged relationship to the data they have generated, ‘being there’ is not the final arbiter of the adequacy of understandings, although a sufficient understanding of proximate contexts is a crucial component of analysis. Issues of context extend well beyond proximate knowledge of settings and individuals. One of our concerns, as secondary analysts, is to seek a critical understanding of how data relate to the broader conceptual questions in which we are interested (see Irwin and Winterton, 2011a for a fuller discussion). Social science researchers are well aware that data carry with them the imprint of how they were generated. Working across data sets allows us to see very immediately the ways in which different project designs and methods generate very different kinds of evidence even when, on the surface, they are exploring similar questions. Bringing data into comparison can offer insights about their specificity and their contextually embedded nature, an issue which needs confronting if we are to build knowledge across differing data sources.
We develop our arguments through three examples. In the first one we take an arguably top-down view of context yet it is one which can be important to qualitative research. This is the issue of how specific qualitative samples and evidence are situated in respect of heterogeneity across relevant wider populations. We consider one Timescapes project where we have used quantitative data as a vehicle for explicating what qualitative evidence is being taken to ‘stand for’. It also provides a systematic basis for building comparisons within a qualitative data set which map onto wider patterns of heterogeneity. There are important issues here for producers and (re-)users of data to reflect on. How do we situate qualitative data? What do we take evidence to stand for? How do we relate the specific and the general? (cf. Ward Schofield, 1993).
Our next examples are grounded in more inductive forms of data analysis. In both we can see how the contexts of research projects frame evidence in specific ways. A strategy of working across data sets requires we understand how this contextual framing is working. Indeed comparative working across data sets soon sheds light on the ways in which data are embedded in the contexts in which they are produced. The second example, within the article, is of active internal collaborations within Timescapes where teams brought their qualitative data ‘into conversation’ in order to explore common themes. Here, the primary researchers became secondary analysts in discussing their data in face to face meetings. The comparison of selected data became an opportunity for primary researchers to ‘hear silences’ in their own data, and encouraged a new interpretation of aspects of their data. Our third example describes a more conventional strategy of secondary analysis, where two of the authors explored archived data not of their own making, and sought to work across several project data sets. Here we explored responses to some questions asked in common across the Timescapes projects. We describe how we tackled this task, and how we began to refine our initial, exploratory, research question. 2 We draw out some implications relating to the importance of understanding context in data production and analysis. We briefly consider the evolving focus of our early analysis: the salience of norms and contexts of lived experience in how research participants frame their accounts of biographical change. We show how we evolved working hypotheses to test out across data sets. Across all our examples we explore some implications of working across Timescapes project data sets, examine the bases on which we can build meaningful comparisons, and discuss challenges as well as opportunities for knowledge building. 3
Situating qualitative data with reference to quantitative evidence on diversity
There has been a resurgence of interest in developing the joint use of qualitative and quantitative data. 4 While joint designs are not the remit of the Timescapes longitudinal study, there are some opportunities for working across data sets. These have been more developed in the Young Lives and Times (YLT) study, which included the generation and use of a range of qualitative and quantitative data. We will focus on a particular example of working across different kinds of data, specifically to show how we might usefully situate qualitative data with reference to survey evidence, with a view to enhancing analysis. The Young Lives and Times study is a qualitative longitudinal study, undertaking mixed qualitative methods with teenagers who were aged 13 and 14 at the outset of the research. They were interviewed, and took part in various visual elicitation methods, initially in 2007 and again in 2008. A 3rd wave of interviews was undertaken in 2010. The YLT study was designed to explore young people’s personal relationships and identities, their experiences of family, friendship and intimacy, their perceptions of biography and time, and their expectations and aspirations for the future. At the start of the project a questionnaire survey was undertaken in the education authority area, and questionnaires were completed by 489 pupils across 10 schools. The year group was chosen in part to parallel the year group being followed by the Longitudinal Study of Young People in England (LSYPE). One of the present authors designed and ran the survey, and was involved in the primary research team. The example raises some interesting questions about if and how qualitative researchers (primary and secondary) can make, and facilitate, links between qualitative and quantitative evidence.
Within the YLT survey questionnaire a number of the questions parallelled ones asked in other surveys to allow points of comparison with other data sets, for example with the national LSYPE. This allows us to check the extent to which the evidence from a relatively small survey corresponds with that of a large national representative survey. If it does we can be more confident in claims based on new lines of enquiry allowed by the YLT survey (for details about the survey and how it aligns with national level evidence see Irwin, 2009).
In the following example we consider a strategy for bringing together data from the YLT survey and the qualitative study. Although an ideal strategy might be an iterative one in which qualitative and quantitative research proceed in sequence, providing a mutually informative resource for successive rounds of fieldwork, the YLT study resources and timetable only allowed for a survey to be conducted alongside the first wave of qualitative research. Both parts of the study drew on separate and dedicated qualitative preliminary research. Participants in the main YL project qualitative research were invited to complete the survey questionnaire. Concomitantly we can see how they are situated on key criteria of interest, with reference to their distribution across a wider population. The substantive issue in this example is the question of how young people’s educational expectations are shaped, and what addressing this can tell us about current social inequalities.
In respect of attitudes to school work a variable derived from responses to three questions provides an index of academic motivation. We can ask: what is the link between academic motivation and expectations about going to university? All things being equal we would expect current motivation to be a major predictor of such expectations. However, all things are not equal. For example, within the YLT survey, parental occupational class, or parental higher education participation are as strong as predictors of children’s expectations as is academic motivation. A simple three way contingency table reveals the relationship between academic motivation and strength of expectation about applying for university, controlling by parental background in higher education.
The data in Table 1 allows us to situate YLT qualitative study participants’ data directly with reference to the distribution along these particular dimensions, manifest in the YLT survey, through their responses to the relevant survey questions. The table is ‘peopled’ with names of a handful of qualitative study participants for illustration. One of the authors has developed a discussion of the substantive issues elsewhere (Irwin, 2009). Here we focus on some of the methodological issues raised by this strategy of representing and organizing data. It has echoes with some mixed method sampling strategies, where a numerical distribution is used as a sampling frame. Often this has been done with reference to ‘objective’ (e.g. socio-demographic) criteria only but there are theoretically productive reasons for doing so with reference also to subjective criteria (Irwin and Elley, 2011). Here we are situating our participants in the qualitative study with reference how background, motivations and expectations relate to the patterning of responses across a broader population. In this way we orient to a more case based, comparative analysis, where cases can be situated in respect of aggregate patterns, as well as relative to one another. For example, we can pinpoint whether our participants manifest typical or atypical expectations, given their backgrounds and academic motivation. We can then ask: how does the qualitative evidence help us to understand their expectations? Thus situating qualitative evidence with reference to patterns revealed in the survey data allows us to make transparent what, in qualitative analysis, we are taking participant cases to stand for. Additionally the strategy formalizes the bases on which we might make comparative qualitative analysis across individual participants. It encourages a consideration of diversity within, as well as difference across, categories. There are potential criticisms of such a strategy. In the example here, in order to map onto survey data, we are imposing an analytic framework which may close off ways of interrogating the qualitative data, or detract from their richness. However, it is not exclusive of other strategies and if it encourages reflection on how we make connections between individual level experience and general evidence on pattern and/or broader social processes then it has served a positive purpose.
Young people’s expectations of going to university, YLT survey data and examples of qualitative study participants, showing how they are situated with reference to the survey distribution on three variables
Details of the derivation of variables are available in Irwin (2009)
Reframing original data through inter-project secondary analysis
Collaborative cross-project work for the purpose of secondary analysis is integral to the design of the Timescapes programme. So far this has been organized in two main ways, in whole project residential meetings and in visits between pairs of projects. In what follows we focus on the process of collaborative analysis through paired project meetings, and some conceptual and empirical outcomes. First, however, it might be useful to consider to what extent this type of collaborative analysis is appropriately described as secondary analysis. We would argue that it is. What may be a significant difference, with respect to secondary analysis of archived data, is that both sides of the pairing we go on to outline had the benefit of being able to explore with the original researchers issues of context, and their intentions and perspectives. As others have suggested (Hammersley, 1997; Mauthner et al., 1998) such information is constitutively integral to data and more than simply background to a project. Like Hammersley we would argue that data are ‘constructed or produced in the course of research’ (Hammersley, 2010: 7).
The direct questioning of the original researcher is not unknown in secondary analysis. Dennis Marsden and Peter Townsend were both interviewed about their original studies, of ‘lone mothers’ (Marsden, 1969) and residential care (Townsend, 1962), by later researchers who used their archived data (Evans and Thane, 2006: Johnson et al., 2010). However, such exchanges could not be classed as pairings, more as consultations with the primary researchers. What the Timescapes experience enabled was a dialogical interrogation of the data and the accounting of a negotiated and unfolding process of sharing and developing understanding. As in any secondary analysis or re-use of data the results were new interpretations and new considerations of the original material, as we will go on to illustrate. In this way, inter-project secondary analysis contributed to asking new questions and the identification of new evidence as the original data were engaged with comparatively.
In what follows we draw out an example from one of three pairings initiated by ‘The Oldest Generation project’ (TOG). A visit to ‘Masculinities and Risk: Transition in the Lives of Men as Fathers’ (henceforth The Fathers Project), at the University of Cardiff, in July 2008 resulted in the identification of emergent themes in both data sets which had not been included in the primary analysis of the TOG data.
Some brief background details will help to outline key features of the projects. TOG data derived from two rounds of interviews with twelve people aged 75 and over at an 18 month interval, with diaries of their lives during that eighteen month period kept by a close family member. The families were spread over the UK and were recruited through the Open University network. At the time of the pairing, only one wave of interviews had been completed by TOG. This first wave of interviews had taken a life history approach. The Oldest Generation has been concerned with the dynamic nature of older people’s relationships and identities in the context of changing structures of intergenerational support. The aim was to explore how and why certain family relationships endure or change over time and how these processes affect the lives of members of the oldest generation. Data from The Fathers Project came from a first round of interviews with first time fathers in 1999 who were revisited at intervals and then eight years later. This sample was then extended to include a group of men from South Wales each of whom was interviewed three times over an 18 month period, again focusing on first-time fatherhood. The Fathers project had a central concern with the extent to which, and how, becoming a first-time father changes identities, relationships and lives over time.
A theme emerged in discussion with The Fathers Project which led to a reviewing of TOG data in a way that had not previously been considered. The theme, ‘discourses of planning and choice’ was explicit in The Fathers Project as interviewees talked about choice in decisions about becoming a parent, highlighting the impact of its timing on their own lives. Fatherhood was presented as something which was planned and its timing weighed up in relation to its impact on their lives. By contrast, the men who were interviewed for TOG, with one exception, made no mention of the timing of becoming a father and that this might be planned or managed in some way. Their stories relate courtship, marriage, shortly (sometimes very shortly) after which they began to have children and then after a few years completing their families. The difference between the two groups of men seems to be that in the Men as Fathers project becoming a father was presented as something which could be deliberated separately from decisions relating to partnering. In contrast, the TOG interviewees presented fatherhood as following from marriage and bringing life changes with an inevitability that excluded choice. As explained by one interviewee:
By the time we were married my party days and celebration days were over, you know, I felt that my obligation was different from – you know, I stopped smoking, didn’t go to pubs. My family become my obligation, sole obligation. And that’s how it stayed ever since, you know. Even though they’ve left home and it’s still this family situation.
Choice and agency plays a lesser role in participants’ accounts, an issue which may relate to their experience of following more structured life course trajectories, or which may relate to the different perspectives of older and younger men in different project contexts. We return to this question later. For the TOG researchers, however, the absence of choice in their participants’ accounts helped formulate new lines of enquiry.
Noting the difference between the two data sets also opened up questions relating to the focus of the two sets of interviews, including gender issues in interviewing, the history of birth control, the effect of retrospection from a late life position, awareness of cohort and generational expectations as well as socio-historical contexts for the evaluation of risk and choice. These had both methodological and substantive implications for the TOG project, for example social and gender differences were in operation at the time of the interview given that interviewees were all over 75, and at least 20 years older than their university based woman interviewer. There is also the likelihood that self-censorship on the part of the interviewees or subjective interviewer assessment of subjects such as birth, pre-marital sex or family limitation meant that certain topics were felt to be off limits in a project where other family members were known to be taking part. Viewed from late life, the timing of a life change is presented in a way that suggests a generational difference which would not have become apparent without the possibility of comparison with data from The Fathers Project. Interestingly, it was in acquiring the status of becoming a grandfather where the effect of the younger generation’s exercise of choice became evident. One interviewee complained at having to wait to become a grandfather:
Oh I was delighted, as a matter of fact I was getting ready to adopt grandchildren. I says ‘what is the matter with my sons, are they not having any family?’
Considering both sets of project data together enabled the emergent theme of planning and choice to be highlighted in ways that had not been evident previously. Where choice was mentioned in The Fathers Project, its significance changed in light of its absence in TOG. Thus where planning and choice had not originally been a focus for interest, a search for its latency or its absence led to the framing of new questions about becoming fathers. This suggested a critical approach to the balance of the TOG interview data where fatherhood was identified as one among many expected life events and transitions, of great significance, but not expressed as a process of active decision-making. It thus became interesting to speculate to what extent this was an effect of a view from the perspective of later life or part of an historical shift in contexts of choice.
Exploring data across TOG and The Fathers Project highlighted the value of project team collaboration within Timescapes. Opportunities to engage in a dialogue between two data sets meant that the primary researchers 5 could critically re-consider their data, noting ‘silences’ or alternative conceptualizations as they compared interpretations of their two data sets. The result was an analysis which was a re-framing, leading to new questions being asked of the original data, for example with respect to choice and awareness of shifts in narrative strategies in relation to time in the lives of interviewees. (Additionally, in the case of TOG, a decision was taken to introduce a more strongly focused future orientation in the subsequent wave of interviews [Bornat and Bytheway, 2010]). The practice of secondary analysis implies a re-visiting and a ‘re-constitution’ (Hammersley, 2010) of the data with opportunities to re-read with the aid of alternative approaches new to the original researchers. The example given here shows how this process worked and with what methodological and substantive outcomes for the research being undertaken by both projects.
Working across Timescapes project data sets and responses to common questions
Introduction: common questions as a point of connection?
In this section we turn to examples of analysis across data sets where the researchers have had no involvement in the design of projects on which they draw. We are seeking insight from data that have been collected by different groups of researchers to address their own specific research aims. All Timescapes projects have common and broadly overlapping interests around identities, relationships and life course dynamics, but differing research questions and strategies. There is a wealth of opportunity to explore various issues in a ‘ground up’ way, exploiting common areas of interest. This undertaking is central to the Timescapes secondary analysis project, underway at the time of writing. In addition, as a way of exploring points of connection across studies, all projects included in their interviews some Timescapes-wide questions, on perceptions of biographical, and historical, changes, and age and generation. The asking of ‘common questions’ therefore provides a direct point of comparison across projects. As anticipated such comparison is far from straightforward given the way questions are embedded in quite different project contexts. We focus on a question about turning points and significant events, intended to generate reflections on biographical change among participants. We took this as our initial entry point into diverse data sets, as a vehicle for starting to familiarize ourselves with the data and as a potentially interesting exercise in its own right. We describe our practise: how we went about doing the analysis, and how it evolved, and seek to identify some general methodological pointers for secondary analysts.
Working across the Timescapes data sets
Thomson and her colleagues have developed the concept of critical moments as a focus through which to explore the links between individual biographies and social structural inequalities (Holland and Thomson, 2009; Thomson et al., 2002). A crucial dimension of their analysis is the ability to move between narratives and very detailed knowledge of participants’ circumstances and lives through time. While this sort of analysis is possible in principle with Timescapes data, it is difficult for secondary analysts to reconstruct the depth of knowledge held by primary researchers. For the secondary analysis team the early process of undertaking analysis across Timescapes projects’ data offers a different kind of entry point, facilitating a comparative cross-project mode of working. This generates difficulties but also invites reflection on how we can build knowledge across diverse qualitative data sets.
Data from the common questions were embedded within larger interview transcripts that were deposited uncoded within the Timescapes Archive. The time required to access relevant data is multiplied when working across several projects. In developing an exploratory analysis we needed to choose how to enter into the data sets. Initially we searched thematically across participants and across projects aiming for breadth. We searched within documents for relevant passages where questions had been asked about turning points or significant events, and then read whole or extensive parts of transcripts to build a picture of the individual participants. In so doing we generated some conceptual ideas, or loosely formulated hypotheses, which we discuss below. A thematic analysis (where we read ‘across’ participants grouping data by theme) risks disembedding data from the contexts in which they were generated. This applied both to the ways the questions were asked and the interview and project contexts in which they were answered. We supplemented our initial thematic review of several projects and their participants with a fuller reading of individual transcripts, across participants, and across waves, on a project by project basis.
It is by developing a detailed knowledge of participant cases that the impact of different interview contexts became evident. An awareness of the manifold ways in which context shapes the data which are produced is particularly crucial for secondary analysts, who will have a limited ‘feel’ for proximate contexts of primary data production. We initially consider how one question was formulated and embedded differently, across projects. The process of comparison across data sets helps to show differing project contexts in sharp relief.
The way in which the Timescapes common questions were asked varied across projects, embedded in the corpus of the interview task in some, and turned to as an ‘add on’ to the main interview in others. Participants had been given advance notice of the questions in one project, and not others. In respect of the question on biographical change, some interviewers asked about turning points, in others for a single turning point, and elsewhere, interviewees were asked simultaneously about significant events or turning points. These phrases themselves orient interviewees differently. Important life course events might be simply named, for example, while, as we will see, turning points was taken by some as a prompt to talk about a point of redirection in their life. While qualitative researchers adapt and tailor their questioning to the interview contexts, it is always important to reflect critically upon how even minor changes in questions and prompts can engender very different kinds of account.
Generating hypotheses for exploring and testing across different project contexts
Here we briefly review how we evolved some different, somewhat loosely formulated, hypotheses from our reading of the data. We go on to suggest how we can make a virtue of difference across the projects since it allows for an exploration of such hypotheses across diverse contexts.
The first review of cross project data on turning points and significant events, based on a consideration of four Timescapes projects, produced data on many kinds of transformations that were described by participants. The kinds of transitions were varied, and often anchored in routine/expected life-course stages, e.g. the movement from school to employment, from home to university, from single to married and childless to parent. These were often described in a very routinized way as a simple list in direct response to the question. Perhaps this reflects a sense of a ‘normal biography’, one where experience is in broad alignment with wider structures and expectations (there is an example of this in the earlier discussion of The Oldest Generation study participants). Across projects, when interviewees were asked about turning points, some of them engaged in detail with the question and talked at length about a redirection of their lives. Where this was the case, typically, there was a linked and extremely strong emphasis on agency, and interviewees revealed the importance to them of the capacity to act to change their lives. In exploring the responses to the question, we began to conceptualize the evidence as offering a potentially interesting lens on personal agency and the contexts in which this is expressed by participants.
Turning points is very ambiguous as a term, so participants’ interpretations varied. However, as we might expect, a turning point appears to have been interpreted by some participants in terms of a biographical re-direction; and it was more commonly identified as such by women than by men. Additionally, such women also link it closely to change in their family roles and relationships.
We illustrate these themes as manifest in the accounts of some of the 13 women interviewed as part of the Work and Family Lives project. The project focus was with working parents and their primary school aged children, and how they experience and manage issues raised by working parenthood. One adult female interviewee described her shock when her mother died in her early 60s, and said:
it pulled me up short and I thought ‘no, life’s too short, I’m in an unhappy marriage, what have I achieved? Nothing. So I instantly changed track.
She then recounted how she decided to go on a college course, gave her husband ultimatums, and then got a divorce, ‘I just steamed ahead and had a plan of what I was going to do’.
Another described her separation and divorce and said she became much more confident and able to manage and that this was:
. . . quite a major point in saying ‘right, I can do stuff . . . if I can do that I can do other things . . . and if I achieve anything, oh I can do most things if I set my mind to it.
Indeed in the case of all four women who appeared to emphasize as a turning point their capacity to act, this involved an act of separation or divorce. Our interest in the manifestation of agency then we might hypothesize as being centred around some crucial life event; perhaps a disruption to a ‘normal biography’. 6 We might hypothesize that where people deviate in some way from what they might have expected previously, where they identify a turning point, this is necessarily accompanied by an expression of self efficacy and agency in enacting a change in their lives. We might then ask, how can we explore this emerging hypothesis in other contexts?
To explore this avenue of enquiry further we turn to another data set. In the accounts of participants in ‘The Oldest Generation’ study, there is a less vivid picture of agency. Again, women commonly talked about turning points in terms of relationships with close family members. One interviewee said:
. . . I suppose, once (my husband) died well then, changes, you have a life before marriage and a life after . . . You’ve got to continue and your children are happy if you are happy.
Another interviewee, asked about turning points, referred to her husband’s death as the biggest thing in her life, and referred to her two grandchildren, then aged 1 and 4, as ‘bringing her through’, that is by taking her mind away from it.
Some older people, then, respond to the question on turning points with accounts which hinge less on personal agency and more on experiences outside their personal control. They adapt to these and describe agency, but their accounts show how they see such agency as being embedded in circumstances. Interestingly, both are linked very closely to inter-personal (familial) relationships. These participants appear to describe a very different picture of agency than do the younger women cited from Work and Family Lives. Perhaps this simply relates to experiences of bereavement. However, it may relate to wider sets of circumstances in which older participants feel they were less able to be authors of their own lives. Or, with very differing implications, it may be that the perspective of later life offers a more sociological view of lives in context: for these older participants distance (and sometimes death of close others) puts agency in context.
In the different studies we have very different accounts of agency which raises a series of questions. Are the different kinds of account evidence of historical changes in people’s ability (real or perceived) to effect change in their lives? Conversely, does such difference relate to the ways in which participants have been oriented within the context of the study? For example, it is plausible that the focus in the Work and Family Lives study encourages participants to foreground their own personal strategies and agency, in contrast to those in The Oldest Generation, a study centring on oral history, in which participants may orient more to their lives in social and historical context. Are there ways in which we can test out alternative interpretations with reference to the extant data sets? We might generate hypotheses which make a virtue of cross project working. For example, one might next seek out accounts by TOG participants of where they experienced or effected non-normative familial or life course transitions. Do they frame agency and self efficacy in a different way in such instances, and how does this compare to the Work and Family Lives interviewees, as cited above?
To sum up, for us as secondary analysts, turning points became interesting as a lens through which to explore people’s accounts of agency in their own biographies. Our evolving thinking on the topic engendered different areas we might seek to follow further, and hypotheses we might explore across data sets. The initial exploratory question evolved into a more clearly defined (although still quite broad) line of enquiry. We saw accounts of turning points as a lens on people’s perceptions of choice and constraint in their immediate relationships. A strategy to develop the analysis further would be to examine and develop this conceptualization more thoroughly within projects and then take emerging hypotheses and test and refine them through further cross-project comparative analysis.
While the example we have given is truncated we hope simply to show that it offered us a ‘way into’ diverse and very extensive data sets; but also that it facilitated forms of engagement with social processes amenable to exploration across projects. It is very obvious that we cannot trawl for data in response to the same questions across Timescapes projects and expect we are simply comparing like with like. We need a sufficiently nuanced understanding of meaning as it is generated within specific projects. For some, this specificity of qualitative data calls into question the possibility of effective secondary analysis or generates insuperable challenges. Working across a diverse set of projects does present real challenges, not least the ways in which meaning is embedded in diverse contexts, forms of question wording, and differing interpretations of the question. Further, participants orient differently to what they perceive as the research aims, and accordingly they attribute relevancies in the accounts they give. We have argued that there is, nevertheless, scope for effective working across data sets, building working hypotheses or conceptual ideas about process which we can then explore and refine through cross-project comparative working.
Conclusion
Context is widely discussed in the literature on secondary analysis, but it is often treated as an issue of proximate knowledge of participants and of their immediate circumstances. We have sought to address different dimensions of the multi-faceted question of context. Through two of our examples we have considered the ways in which data are embedded in, and shaped by, different project aims, methods and modes of questioning. Through all three of our examples we have shown how strategies of comparison can be productive, both for analysis and for conceptual development. In the first example we illustrated the potential value of working across qualitative and survey data, using the latter to better understand how participants in a qualitative sample are situated in respect of diversity across the population. We did so with reference to socio-demographic, and attitudinal, indices. Ordering qualitative data in this way also facilitates forms of internal comparative analysis made with reference to wider social patterns. Second, we considered the practice of bringing together primary project teams in paired meetings, where data were brought into conversation and points of comparison helped identify differences and ‘silences’ within data sets which were interpreted as analytically meaningful and gave rise to new questions and avenues of enquiry. Finally we considered the scope for working across a range of qualitative data sets where some questions had been asked in common, but across quite differently conceived projects. This shone light on the contextual specificity of project data. We developed some working hypotheses and conceptual ideas relating to expressions of agency, biographical change and turning points. We suggested that such hypotheses could be evolved, rejected or refined by a strategy of comparison across different projects and contexts. While embedded in the objectives of Timescapes, we believe that these strategies are all ones which might be helpfully reproduced elsewhere.
Footnotes
This research was funded as part of ESRC Timescapes, RES 347250003.